Behind New US AI Diffusion Rules: Geoeconomics & Geopolitics

In the last week of its tenure, the Biden administration passed a legislation called the AI Diffusion Rules. It has two stated objectives: First is to ensure US leadership in the AI technology stack, and second is to restrict geopolitical adversaries (read China) from accessing cutting-edge technologies.

The contours of the rules are complex and pose major ramifications for technology or economic development, with major geopolitical implications.

Understanding AI Diffusion Rules

The rule demarcates the world into three categories: Tiers 1, 2 and 3. Tier 1 countries face no restrictions in accessing US-developed technologies, Tier 2 countries face some restrictions in the form of licensing, and Tier 3 countries are red-flagged and banned. 

Restrictions have different forms. Any company in a Tier 1 country gets full access to advanced AI chips and development modules. 

But if it wants to locate part of its operations in a Tier 2 country, it must ensure that 75 per cent of its total compute capacity is in a Tier 1 country. Additionally, no Tier 2 country can host more than 7 per cent of the total compute capacity of the company. 

Broadly, compute is a metric of measurement that refers to the number of calculations that a processor can do in one second.

A non-US company, looking to set up operations in a Tier 2 country, has to apply for national validated end-user (NVEU) authorisation with restrictions in the US. It can only 'deploy' up to 1,00,000 Nvidia H100-equivalent AI chips (latest AI accelerator chips) by the end of 2025, 2,70,000 by end-2026, and 3,20,000 by end-2027.

Getting an NVEU is expected to be more difficult for a company based in a Tier 2 country.

Validated end-user (VEU) authorisation allows US exporters to ship designated items to pre-approved entities under a single general authorisation, instead of multiple individual export licenses.

The limitations are not only restricted to hardware, but also on how much of the AI model can be deployed in Tier 2 countries. Tier 1 country-based companies can deploy AI chips with some restrictions, but Tier 2 entities face a blanket ban on access to compute capacity from Tier 1-country based companies.

India In Tier 2: What It Implies

Ironically, despite the US-India Initiative on Critical Emerging Technologies and expanded global partnership in the field of developing dual-purpose technologies, India finds itself in Tier 2 list.

This has a twin impact on India’s AI capacity development.

First, it severely restricts India’s ability to acquire and develop its own hardware infrastructure to develop its indigenous AI capacity. Second, there will be an impact on India’s app development, as India’s current focus, a bit surprisingly, is on building its expertise more on the application layer rather than the hardware.

The restriction effectively means that Indian companies cannot develop their AI models by leveraging Tier 1-owned data centres. For instance, access to Meta’s Llama or OpenAI databases, which form the base on which all India-based AI start-ups are working, now face the risk of being severely restricted.

It also means that global majors like Meta or Microsoft (which are driving the global AI development) will restrict their own AI development operations in India. That will impact the capacity building of the Indian AI ecosystem.

Indian talents, who intend to work on AI, will have to leave Indian shores to join Tier 1-based companies at their home locations, leading to further brain drain.

Even though it may come as a heartbreak for those looking forward to new chapters of cooperation between the US and India, the fact remains that this is not the first time India has faced a technological embargo from the USA.

Earlier, India faced embargoes on nuclear and space technologies. However, the nation managed to develop indigenous capacity in both fields. Perhaps that is the inherent reason why India finds itself on the Tier 2 list. The US administration probably still perceives India as a defiant state.

However, a more pressing question currently is whether India can bypass this regulation to develop its AI capacity or not.

What It Means For Tech Biz

Ironically, the rules were slipped in by the Biden administration literally days before the Trump administration took office. In a sense, this may have been done to restrict any potential ‘damage’ Trump 2.0 could do to American strategic business interests.

However, American businesses may still suffer. For many American companies which have already invested (or have commitments in the pipeline) to develop their capacities in Tier 2 countries like India, the rules will mean a major commercial setback. 

Mandates to locate 50-75 per cent of total computing capacity in Tier 1 countries is a costly proposition, and negate the reason why they shifted some of their operations to Tier 2 countries in the first place.

Businesses like Nvidia and Microsoft, at present very cosy with the Trump administration, have heavily criticised the rules.

Ned Finkle, Nvidia’s Vice-President, openly criticised the rules in a public blog, in which he accused: “In its last days in office, the Biden administration seeks to undermine America’s leadership with a 200+ page regulatory morass, drafted in secret and without proper legislative review.”

He argues that while the rules are cloaked as ‘anti-China’, they actually do precious little to safeguard American interests, as it risks weakening America’s global tech cost competitiveness.

Finkle is joined by Brad Smith, Vice-Chair and President of Microsoft, who also wrote an open blog. He went a step further: “As drafted, the rule undermines two Trump administration priorities: Strengthening US AI leadership and reducing the nation’s near trillion-dollar trade deficit. Left unchanged, the Biden rule will give China a strategic advantage in spreading over time its own AI technology, echoing its rapid ascent in 5G telecommunications a decade ago.”

Smith argues that if the US restricts AI technology access to Tier 2 countries, they will be forced to look elsewhere (again, read China), as no country can afford not to develop their own AI capacity.

He recognises that the rules categorise many US allies and partners in Tier 2, which is unfair. He specifically writes, “This includes many American friends, such as Switzerland, Poland, Greece, Singapore, India, Indonesia, Israel, the UAE and Saudi Arabia. These are countries where we and many other American companies have significant data centre operations.”

The launch of DeepSeek by China around the same time the rules were passed, has shaken a hornet’s nest. Some in the US feel that DeepSeek proves such stringency should have been adopted much earlier. 

Others argue that DeepSeek highlights the futility of such regulations, especially at this stage, as it only gives China more space, while restricting US businesses from spreading. 

What Will Trump Do?

Technically, the rules become operational after 120 days of adoption. What will the Trump administration do about it?

Experts feel the rules create a sense of mistrust among US allies and risk alienating some. Some partner countries will be forcibly restricted behind the frontier of future technology. This may also widen the divide between the first world and the third world. 

The USA, already sharply polarised internally, may now face more global turmoil among its allies and trade partners. However, the Trump administration may love this regulation as it offers a leverage for negotiations.

For example, the criteria to categorise a country Tier 1 or Tier 2 is not well-defined. This alone can be a wonderful tool for carrot-and-stick diplomacy. Permits to NVEU can be a tap that can be controlled to align foreign countries with the economic or political hardlines of the US. 

The Trump administration may actually project the rules not as a divisive regulation, but as a tool for streamlining ‘global order’. In such a scenario, the new US administration may amend or at least implement this regulation differently, but is very likely to maintain policy continuity.

(The writer is a New Delhi-based economist with over a decade's experience in studying the digital sector. Views are personal)